stringr str_replace_na() in R: Replace NA Values in Strings

The str_replace_na() function in stringr replaces NA values in a character vector with a string label. The default replacement is the literal text "NA", but you can pass any single string (often "", "missing", or "unknown") to get NA-safe output for paste, concat, and display.

str_replace_na(x)                         # default replacement "NA"
str_replace_na(x, "missing")              # custom label
str_replace_na(x, "")                     # treat NA as empty string
str_replace_na(c(1, NA, 3), "0")          # coerces input to character
paste(str_replace_na(x), "!", sep = "")   # NA-safe paste
str_c("hi ", str_replace_na(x, "guest"))  # NA-safe str_c
tidyr::replace_na(df, list(name = "?"))   # data-frame equivalent

What str_replace_na() does in one sentence

str_replace_na(string, replacement = "NA") substitutes every NA element in a character vector with the chosen replacement string. Unlike str_replace(), it does not use regex; it only swaps NA for a string. Non-character input is coerced to character first.

This matters because base R's paste() and stringr's str_c() propagate or preserve NA in surprising ways. str_replace_na() is the one-step fix that makes a vector safe for concatenation and display.

Syntax

str_replace_na(string, replacement = "NA") takes two arguments: the input vector and a single replacement string.

The literal string "NA" (four characters: N, A) replaces each missing value. The result is a regular character vector with no NA left.

Four common patterns

1. Custom label for missing values

Pass any single string. Common choices: "missing", "unknown", "n/a", "--", or "" for blank.

2. NA-safe paste

paste0(NA, "!") returns the string "NA!", which is rarely what you want. Replacing NA with "" first gives clean output.

3. Coerce non-character input

The numeric vector is coerced to character before substitution. The output is always a character vector; use as.numeric() after if you need numbers back.

4. Empty string for CSV or table output

For CSV exports or printed tables, an empty cell often reads cleaner than the literal text "NA". This pattern is also useful for joining downstream string fields.

str_replace_na vs replace_na vs coalesce

Three functions handle NA replacement in R, each at a different level.

Decision rule: working on one character column, use str_replace_na(). Working across many columns of a data frame, use tidyr::replace_na(). Replacing NA with values from another vector (not a constant), use dplyr::coalesce().

Common pitfalls

Pitfall 1: confusing the string "NA" with the value NA. After str_replace_na(x), the result contains the literal characters N and A. Tests like is.na(result) return FALSE everywhere; tests like result == "NA" return TRUE at the replaced positions.

Pitfall 2: trying to use a regex pattern. str_replace_na() does not take a pattern argument. It targets only NA values. If you want to replace NA AND a regex match (like empty strings), chain str_replace_na() followed by str_replace().

Pitfall 3: factor input loses levels. Calling str_replace_na() on a factor coerces to character, so factor levels are dropped. If you need the result back as a factor, wrap the output in factor() with the original levels plus the replacement label.

Try it yourself

Related stringr functions

After mastering str_replace_na, look at:

• str_replace(), str_replace_all(): replace regex matches in non-NA strings
• str_c(): NA-aware concatenation; pair with str_replace_na() for safe joining
• str_detect() with is.na(): locate NA elements before transforming them
• tidyr::replace_na(): data-frame-level NA replacement across many columns
• dplyr::coalesce(): vector-level NA fill from a fallback vector

For full reference, see the tidyverse stringr documentation.

FAQ

What does str_replace_na do in R?

str_replace_na(x) returns a character vector where every NA in x is replaced by a single chosen string (default "NA"). It is part of the stringr package and operates only on character vectors (non-character input is coerced first). The result has no missing values; is.na() returns FALSE for every element.

What is the difference between str_replace_na and tidyr replace_na?

stringr::str_replace_na() works on a single character vector and accepts one replacement string. tidyr::replace_na() works on a data frame or list and accepts a per-column replacement via a named list. Use stringr inside mutate(col = str_replace_na(col, "...")); use tidyr when replacing NA across many columns at once.

How do I replace NA with empty string in R?

Call str_replace_na(x, ""). This converts each NA element to the empty string "", which renders as a blank cell in tables, CSV exports, and concatenated output. The output is a character vector with the same length as the input.

Does str_replace_na work on numeric vectors?

Yes, but the result is always a character vector. The numeric input is coerced via as.character(), then NA values are replaced. If you need a numeric result with NA filled by zero, use dplyr::coalesce(x, 0) or x[is.na(x)] <- 0 instead.

Can I use str_replace_na inside dplyr mutate?

Yes. df |> mutate(name = str_replace_na(name, "unknown")) rewrites the name column in place, replacing every NA with "unknown". This is the idiomatic pattern for cleaning a single character column before reporting or joining.